Apparently, that transformer has ~1.5H inductance. I'll plug that in...
If I simulate that circuit, but replace a 100Meg R (functionally near open circuit) for the 500K R, there's almost no difference in the frequency response. I.E., 500K is way
too much resistance to allow the inductive portion to ever come into play.
Notice how the other (Anderton?) filter uses a 20K "depth" POT? Even that's overkill.
Don't add another POT. Replace the 500K POT with a 10K POT (10K seems to work better with 1.5 Henries than the 5K I mentioned before.)
(There's a real honest-to-God boost here @~1.25KHz. But it would be greater with other, simpler filter circuits. No matter.)
See the next graph? That's the one with the 500K. The series LC filter basically isn't present.
Not significantly different than the 10K graph, other than shifting the initial corner frequency, maybe. And lifting the signal a bit closer to unity.
Your current setup below-- you've got the POT wired as a mixer between the two filters.
With the inductive half "full ON", and the RC half mostly isolated, here's the Freq response. I don't think it's anywhere near as interesting as the original circuit with the 500K removed, or reduced... (first two graphs). There's no secondary "spike", and the higher freq response just tables at -25dB.
With the POT turned the other way, the graph looks like the graph above this one. You're "mixing" the filters alright (introducing the notch), but not very effectively.
Whew. That was epic capture/copy/pasting. Sorry.